AI/ML Group Beam Prediction for Lower-Overhead NR Communication
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
The implementation of group beam (pair) prediction in new radio (NR) systems using artificial intelligence/machine learning (AI/ML) models is not adequately addressed, leading to increased downlink resource overhead, power consumption, and complexity in beam management processes.
Innovation Solution
A wireless communication method involving spatial-domain group spatial filter prediction using AI/ML models, where a terminal device transmits capability information indicating support for a target type of spatial filter prediction mechanism, and network models perform spatial-domain group spatial filter prediction to optimize beam management.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If traditional beam management methods are used in NR systems, then beam management can be performed, but downlink resource overhead increases and system performance is suboptimal
Solution Approach 1:
The patent uses AI/ML models to predict beam pairs by learning from historical measurement data and beam management patterns. Instead of performing complete beam management procedures for every transmission, the system creates predictive copies of beam selection decisions based on trained models, thereby reducing downlink resource overhead while maintaining beam management performance
Solution Approach 2:
The system performs preliminary beam pair prediction using AI/ML models before actual beam management execution. By pre-computing predicted beam pairs based on historical data and current channel conditions, the system prepares optimization decisions in advance, reducing the need for extensive downlink resources during actual beam management operations
2Reliability
If traditional beam management methods are used in NR systems, then beam management can be performed, but power consumption increases
Solution Approach 1:
The terminal device uses AI/ML models to predict beam pairs locally, creating predictive decisions that reduce the need for power-intensive measurement and reporting operations. The model processes historical data to generate beam selection predictions, thereby copying intelligent decision-making capabilities to the terminal side and reducing overall system power consumption
Solution Approach 2:
The terminal device performs self-service beam management by executing AI/ML models locally to predict beam pairs. This self-service approach enables the terminal to make autonomous beam selection decisions without requiring extensive network assistance, thereby reducing power consumption associated with network-terminal interactions and processing
3Reliability
If traditional beam management methods are used in NR systems, then beam management can be performed, but process complexity increases
Solution Approach 1:
The patent copies complex beam management intelligence into AI/ML models that can be trained offline and deployed at the terminal device. These models encapsulate complex decision-making logic for beam pair selection, thereby reducing the complexity of real-time beam management operations while maintaining or improving performance through data-driven predictions
Data Source
AI summary
A wireless communication method includes: transmitting, by a terminal device, first capability information; where the first capability information is used for indicating whether the terminal device supports a target type of spatial filter prediction mechanism, and within the target type of spatial filter prediction mechanism, one or more network models are used to perform a spatial-domain group spatial filter prediction.